Next: Computer simulation methods Up: Main Previous: Electrical conductivity and

Experimental method

Cement pastes were prepared from Type I white Portland cement and deionized water. The oxide content and cement phase analysis are given in Table 1. White cement was used in these experiments due to its high calcium silicate content, making it the best approximation to C₃S without the expense of pure C₃S. The computer model of hydration that was used is based on C₃S-only hydration. Standard cement chemistry notation is used, where C=CaO, S=SiO₂, and H=H₂O. Cement powder was added to the water to obtain a water:cement ratio of 0.5, and then mixed for 5 minutes in a planetary-type mixer at ambient conditions. Fresh pastes for impedance measurements were tapped into the annular region between an inner rod and outer cylinder, both made of stainless steel. A grooved Plexiglass base supported the rod, cylinder and fresh paste. Samples were stored in a sealed plastic container above a pool of 25ºC tap water, and remained in the container at all times except during the data acquisition period. Fresh pastes for pore fluid expression were cast into 51 mm x 178 mm (2 inch by 7 inch) plastic compression cylinders to give approximately 300g samples, while pastes for degree of hydration measurements were cast into smaller, plastic cylinders to give approximately 50 g samples. Both were sealed and stored at ambient conditions.

The bulk paste conductivities, σ, were obtained by impedance spectroscopy [8], which allowed bulk and electrode resistances to be separated. Data were collected with a low frequency gain-phase/impedance analyzer interfaced to a microcomputer. Measurements were made at logarithmic intervals in frequency starting at 11 MHz and ending at 5 Hz, with 20 spot readings per decade. Acquisition of this data took approximately 1 minute. The value of real impedance at the minimum in the imaginary impedance between the electrode and bulk arcs gave the d.c. bulk resistance [1]. Bulk conductance was converted to conductivity by normalizing by the specimen geometry [1].

The pore fluid within the paste was extracted using a steel die according to previously defined procedures [9]. Samples were demolded, crushed, and fluid obtained within a period of 15-30 minutes. For specimens older than seven days, two samples were usually necessary to obtain enough fluid for conductivity measurements. The die was loaded incrementally as follows: 13,000 N/s (3000 lbs/s) up to approximately 445,000 N (100,000 lbs), and then increments of 89,000 N (20,000 lbs), with a 2-3 minute hold at each point to a final load of one MN (220,000 lbs), or approximately 480 MPa (70,000 psi). Pore fluid was captured in 30 ml syringes and transferred to 20 ml polystyrene sealed containers. These precautions were intended to minimize contact with carbon dioxide. Pore fluid conductivity (σ_o) data was obtained within hours of extraction using an in-line conductivity probe.

Degree of hydration α and, ultimately, capillary porosity, were obtained by loss on ignition measurements. Samples of 2-3 grams of cement paste were placed in an alumina crucible and heated at 105º C for a minimum of 4 hours, weighed, then heated to 1000º C for 2 hours. A final weight was obtained when the furnace returned to 105ºC after 8-10 hours. Prior to final set, which took place at approximately six hours, freshly-mixed pastes were placed directly into the crucibles, while after final set had been achieved, hardened pastes were ground to a particle size of less than 600 micrometers before heating. Based upon the above weights, the degree of hydration was obtained by the expression [10]

where W₁₀₅ and W₁₀₀₀ are the weights at 105º C and 1000º C, respectively. Capillary porosity, φ, was calculated from α assuming that cement paste consists of unhydrated cement, hydration product, and capillary porosity. This being true, the relation between these two parameters is given by [11]

where w/c is the water:cement weight ratio, and 3.2 is the specific gravity of cement.